Along with mathematics, computer science has the highest drop out rates in the UK, with one in ten students not continuing into second year. Computer science is notoriously difficult and is often seen as being a subject for which people have a natural ability. Yet recent research (Dweck, 2006) shows that less able students such as girls or African American students, for example, can do better than predicted in subjects such as maths or engineering after specific interventions. This research highlights that a person’s theory about their ability, or intellect, shapes their future success in that particular domain. Those who believe that intelligence can be grown and improved – the growth mindset – tend to persevere for longer, believe in effort, display less helplessness after setbacks and do better on class tests than those who think intelligence is a fixed entity – the fixed mindset. This research project developed three mindset interventions which were used in a first year computer science class. The mindset intervention worked best when it was both taught and integrated into the learning design. Those who were in this group showed a shift in mindset after the intervention and a larger increase in test scores at the end of the academic year, compared to the control group. The findings suggest that small changes can be made to students’ experiences which make a big difference to their motivation and performance.
The project aimed to improve learning on the first year Computing Science programming course by introducing interventions based on Dweck’s mindset work on attitudes to learning and intelligence. This was to be achieved via a mix of staff training on Dweck’s work alongside adjustments to the instruction, exercises and support given to students on the course.
Many students behave as though they believe that ability is a fixed commodity and once it has been shown that they can’t do something then they should give up. Dweck (1999) has shown that our mindset towards ability levels has a crucial effect on our learning. She identifies two categories of learners, those with a fixed mindset (the students described above) and those with a growth mindset, who know that persistent effort and attention to data gleaned from failures will lead to the desired learning. Blackwell et al (2007) have shown experimentally that learners move from fixed to growth mindset with suitable interventions, and that test scores improve over time.
Dweck’s work on mindsets highlights a number of ramifications for learning. Those with a fixed mindset tend to be interested only in performance goals – they feel a need to be seen to be achieving well at all times, since this broadcasts their ability to the world; those with a growth mindset know that they can grow their ability and so set learning goals. They are classical deep learners. Those with a fixed mindset tend to disregard formative feedback, since the very idea goes against their belief of ability being essentially fixed; someone with a growth mindset will use all the data available in order to help them learn. In fact, negative feedback of any kind is likely to lead the fixed mindset learner to give up, to display a helpless response, or to avoid, because it represents an insurmountable barrier to further progress.
The following Dweck-related studies are of particular interest. A study of minority groups at university-level showed that they performed better and were more engaged after learning about mindsets (Aronson et al., 2002). Entering university with a fixed mindset has been shown to affect negatively both students’ self-esteem and grade point average compared with those with a growth mindset (Dweck, 1999). Another study showed that HE students with a fixed mindset completely ignored useful formative feedback after a failure (Mangels et al., 2006). A study of mentors (Heslin et al, 2006) showed that those personally holding a fixed mindset were likely to spend less time with mentees than those holding a growth mindset, further reinforcing the fixed-mindset notion that education is all about showing how much intelligence you have rather than fostering and growing it. Indeed, in a related study (Rosenthal and Jacobson, 1992), teachers were told that pupils in some classes were bloomers (i.e. they had potential), irrespective of their actual ability levels; the pupils in these classes ended up doing better than other pupils of similar ability levels.
Three different Dweck-related teaching interventions were designed.
1. Mindset training
A series of four 10-15 minute teaching sessions were developed for delivery by tutors at the start of their small group, two hour tutorial/lab sessions – to be used over four consecutive weeks early in the term. The tutors were taught about the material for each of these sessions in the week prior to its delivery. Each session involved the tutor talking about a particular aspect of mindsets and then taking the students through a reflective exercise focusing on their own learning experience and relating it to the mindsets work. Tutors were then encouraged to refer back to this material when working with students encountering problems in the lab environment. The four sessions introduced: fixed and growth mindsets; performance and learning goals; responses to feedback; role models and the neuroscience underpinning mindsets. In addition, a short document was created that introduces the fundamental aspects of mindset research.
A crib sheet was created containing a list of questions/hints/pointers for what to do when a student gets stuck. Each item on the sheet linked to a wiki entry explaining the item in much more detail and with examples. Tutors were encouraged to only help students via the crib sheet. So, if a student was stuck, the tutor would work through the diagnostic set of questions on the crib sheet with the student in order to find the problem. The crib sheet was partly inspired by McCartney et al (2007) outlining a collection of 27 ways in which computing students get unstuck.
3. Rubric on making use of feedback
A Dweck-inspired result (Mueller & Dweck 1998) showed that a simple exhortation, based on the science, to make use of the feedback given on assessment tasks, did improve motivation and performance. For this intervention, the following related script or rubric was placed onto feedback sheets attached to the students’ fortnightly exercise submissions: “Remember, learning to program can take a surprising amount of time & effort – students may get there at different rates, but almost all students who put in the time & effort get there eventually. Making good use of the feedback on this sheet is an essential part of this process.”
The design of the feedback sheet involved in the rubric intervention was itself enhanced with Dweck-related aspects. For example, the score the student received was at the bottom of the second side to reduce the focus on the summative feedback. The feedback items they should particularly pay attention to were at the head of the document, accentuating the formative feedback. There was encouragement to read all points, even if not specifically identified for this student, encouraging learning goals generally.
The three interventions were used in a semi-controlled experiment, encapsulating both their delivery and evaluation, described in detail in the next section.
As researchers were unsure in advance which if any of the interventions would have an effect, students were exposed to various interventions according to the tutorial group they were in. In total there were eight different treatment conditions, giving all possible combinations of the interventions from none of them to all three.
The experiment was designed to run for the first six weeks of term, after which point there is a class test. The results of this class test were used to determine whether the interventions were having any effect. The students completed two questionnaires, in weeks one and seven, to ascertain measures of mindset, self-efficacy, hope and positive and negative affect.
A statistical analysis of the results after the intervention showed that mindset had shifted towards the growth mindset in those who had been taught about mindsets (see Figure 1). It is interesting to note that those who were not taught about mindset tended towards a fixed mindset over time.
Further analysis showed that the class tests taken immediately after the intervention and the final degree exam in May showed a significant effect between students who had received both the mindset training and the feedback rubric and their final score. The training and the rubric together were producing an effect size of between 0.5 and 1 grade improvement. (See Figure 2 and 3 below.)
Obvious confounding factors were allowed for. The previous programming experience of the students was measured through a self-report, and gave a hugely significant and sizeable effect. The time of day of the tutorials and their position in the week (and hence in the learning cycle design of the course) were also considered. The result given above is independent of these factors.
A survey of the tutors after three of the four teaching sessions showed that they had in fact delivered the material in very different ways, with some hardly touching on it at all. It was clear that some of the tutors were not fully on board with the mindsets research and were in fact themselves displaying a strongly-fixed mindset. Such attitudes in tutors are likely to affect their behaviour towards their students to a large extent, according to (Heslin et al, 2006). These tutors were given access to a range of further research papers and in particular the neuroscience studies that relate to Dweck’s work, which they found more engaging than our earlier teaching. Additionally, an overview document covering Dweck’s main findings was developed and all tutors were asked to get their students to read this in the final teaching session. In this way it was possible to ensure that all students in the mindset training intervention definitely had knowledge of Dweck’s work.
This issue with the tutors highlights a key aspect of the mindsets research. It is at least as important that teachers/tutors/mentors have a growth mindset as it is for their students. To affect a really sizeable shift in students with respect to their attitudes towards learning, it may be required to instigate a major project to do the same for staff members.
Since both the teaching of Dweck’s findings and the use of a rubric to direct students on the best use of feedback produced strong results in the experiment, both these aspects have been introduced across the class in the current semester. However, the training has been given by the lecturer this time rather than the tutors and it is uncertain whether this is as effective as the teaching sessions given by tutors in small groups. This will be addressed in the next session.
It is also clear that the experimental design had some weaknesses. For example, the application of treatment conditions to students was not truly randomised and so re-running the experiment would be of value.
The materials are available to others on application.
The following papers that support Dweck’s theory of mindsets may be of interest:
Aronson, J., Fried, C.B., & Good, C. (2002), ‘Reducing stereotype threat and boosting academic achievement of African-American students: The role of conceptions of intelligence’, Journal of Experimental Social Psychology
Blackwell, L.S., Trzesniewski, C.H., Dweck, C.S. (2007) ‘Implicit Theories of Intelligence Predict Intelligence across an Adolescent Transition: a Longitudinal Study and an Intervention’, Child Development 78(1), 246-263
Dweck, C.S. (1999) ‘Self-Theories – Their role in Motivation, Personality and Development’, Essays in Social Psychology, Philadelphia: Psychology Press
Heslin, P., Wanderwalle, D. & Latham, G. (2006) ‘Engagement in employee coaching: The role of managers' implicit person theory’, Personnel Psychology
Mangels, J.A., Butterfield, B., Lamb, J., Good, C., Dweck, C. S., (2006) ‘Why do beliefs about intelligence influence learning success? A social cognitive neuroscience model’, Social Cognitive and Affective Neuroscience 1 (2): 75-86
McCartney, R., Eckerdal, A., Mostrom, J., Sanders, K., Zander, C., ‘Successful students strategies for getting unstuck’, ACM SIGCSE Bulletin 39(3), 156-160
Mueller, C.M. & Dweck, C.S. (1998). ‘Intelligence praise can undermine motivation and performance’, Journal of Personality and Social Psychology. 75, 33-52
Rosenthal, R. & Jacobson, L., (1992) Pygmalion in the classroom. Expanded edition. New York: Irvington
To find out more about mindset click here